EP2854288B1

EP2854288B1 - A matching network for use with an amplifier

Info

Publication number: EP2854288B1
Application number: EP14191308.7A
Authority: EP
Inventors: Jonas LINDSTRAND; Carl Bryant; Henrik Sjöland
Original assignee: Telefonaktiebolaget LM Ericsson AB
Current assignee: Telefonaktiebolaget LM Ericsson AB
Priority date: 2011-08-03
Filing date: 2012-07-17
Publication date: 2018-03-07
Anticipated expiration: 2032-07-17
Also published as: BR112014002542B1; EP2740213B1; EP2740213A1; CN103748785A; BR112014002542A2; CN103748785B; US8554162B2; US20130033321A1; EP2854288A3; EP2854288A2; WO2013017407A1

Description

The invention described herein generally relates to matching networks for amplifiers.

BACKGROUND

Current wireless technology trends towards increasing numbers of wireless standards and radio frequency (RF) bands to support wireless communications have led to the development of multi-standard, multi-band cellular systems. Such efforts have produced well-performing wideband receivers and frequency synthesizers. However, power amplifiers having the desired performance, e.g., Power-Added Efficiency (PAE), output power, etc., across the multiple frequency bands remains a challenge for such cellular systems.
While various groups have attempted to solve this problem, the results generally do not provide sufficient efficiency across a wide frequency band, undesirably require multiple amplifier stages, do not provide wideband operation, etc. For example, "A Polyphase Multipath Technique for Software-Defined Radio Transmitters" by R. Shrestha, E.A.M. Klumperink, E. Mensink, G.J.M. Wienk, and B. Nauta (IEEE J. Solid-State Circuits, vol. 41, no. 12, pp. 2681-2692, 2006) provides a wideband solution, but the output power is insufficient and the efficiency across the wideband is low relative to single band amplifiers. "A 1.9 GHz 1 W CMOS Class E Power Amplifier for Wireless Communications" by K.C. Tsai and P.R. Gray (ESSCIRC, pp. 76-79, 1998), referred to herein as the Tsai solution, and "A 29 dBm 70.7% PAE Injection-Locked CMOS Power Amplifier for PWM Digitized Polar Transmitter" by J. Paek and S. Hong (Microwave and Wireless Components Letters, vol. 20, no. 11, pp. 637-639, 2010), referred to herein as the Paek solution, provide alternative solutions using injection-locked power amplifiers. These solutions, however, require multiple amplifier stages and have not shown wideband frequency operation. Another solution, "A 65 ), referred to herein as the Apostolidou solution, provides a wideband solution with improved PAE and output power, but requires multiple amplifier stages, which undesirably increases the chip area and power consumption of the amplifier. US 2007/126505 discloses a method and apparatus for use in power amplifiers for reducing the peak voltage that resistors are subjected to. Figs 6-8 of US2007/126505 show a transformation network, including various inductors and capacitors, which is designed to stabilize the power amplifier. The network may include an inductor and capacitors serially coupled between an output of the power amplifier and the load. "A 2.4 GHz fully integrated CMOS power amplifier using capacitive cross coupling" by J. Young Hong, D. Imanishi and K. Okada, A. Matsuzawa, (Wireless information technology and systems (ICWITS), IEEE International conference, August 28, 2010 pages 1-4) presents a power amplifier with a capacitive cross coupling to solve the voltage-stress issue. In "A fully digital multimode polar transmitter employing 17b RF DAC in 3G mode" by Z. Boos, A. Menkhoff, F. Kuttner, M. Schimper, J. Moreira, H. Geltinger, P. Pfann, A. Belitzer and T. Bauernteind, (Solid-State cicuits conference digest of technical papers (ISSCC), IEEE International, February 20, 2011) a multimode multiband power amplifier is presented using digital polar transmission concept. In "10-Gb/s limiting amplifier and laser/modulator driver in 0.18-/spl ,u/m CMOS technology" by S. Galal and B. Razavi a limiting amplifier incorporating active feedback, inductive peaking and negative Miller capacitance to achieve a voltage gain of 50 dB, a bandwidth 9.4 GHz and a sensitivity of 4.6 mV_pp for a bit error rate of 10^-12 while consuming 150 mW. "A novel LC T-structure filter integrated with CMOS mixer for image rejection" by A-T. Phan, C-W. Kim, M-S. Kang, Y-A. Shim and S-G. Lee (Intelligent signal processing and communication systems, ISPACS Proceedings of 2004 international synopsium on Seoul, Korea, November 18-19, 2004) presents an image rejection mixer in heterodyne architecture for 2 GHz band applicatations based on 0.18 µm CMOS technology.
Thus, there remains a need for an improved power amplifier useful in wideband RF scenarios.

SUMMARY

According to a first aspect there is provided a matching network for matching an output impedance of an amplifier circuit to an external load resistor (R_L). The matching network comprises:

two first reactance units each comprising two inductors (L_D1, L_D2) serially couplable between a signal ground and respective first and second outputs of the amplifier circuit for generating a first positive reactance at low frequencies and a second positive reactance at high frequencies;
two second reactance units each comprising at least one series capacitor (C_s) and at least one series inductor (L_s) serially couplable between a respective end of the external load resistor (R_L) and the respective first and second outputs of the amplifier circuit, for generating a negative reactance at low frequencies and a third positive reactance at high frequencies; and
two third reactance units each comprising at least one control capacitor (C_D) couplable at one node between the respective end of the external load resistor (R_L) and the respective end of the second reactance unit and at the other node between the inductors (L_D1, L_D2) of the respective first reactance unit, for generating a short at high frequencies so as to reduce a parasitic capacitance at the first and second outputs of the amplifier circuit at high frequencies,
wherein a generally constant impedance is provided across a wideband frequency range.

The signal ground may correspond to the supply voltage. According to a second aspect there is provided an arrangement comprising the matching network of the first aspect and the external load resistor (R_L).

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1 depicts a power amplifier circuit.
Figure 2 depicts signal diagrams for the signals at selected points of the power amplifier circuit of Figure 1.
Figure 3 depicts a circuit diagram for an RF current.
Figure 4 depicts a circuit diagram for an RF current generator.
Figure 5 depicts a circuit diagram for an RF current generator.
Figure 6 depicts a circuit diagram for an RF current generator.
Figure 7 depicts a circuit diagram for a matching.
Figure 8 depicts a circuit diagram for a matching network according to an exemplary embodiment.
Figure 9 depicts a circuit diagram for a matching.
Figure 10 depicts a circuit diagram for a matching.
Figures 11A and 11B respectively depict supply voltage and cascode bias vs. input power for the amplifier circuit comprising the components of Figures 1, 3, and 7.
Figure 12 depicts output power, power gain, and PAE vs. input power for the amplifier circuit comprising the components of Figures 1, 3, and 7.
Figures 13A and 13B respectively depict output power and PAE vs. frequency for the amplifier circuit comprising the components of Figures 1, 3, and 7.
Figures 14A and 14B respectively depict gain error and phase error vs. input power for the amplifier circuit comprising the components of Figures 1, 3, and 7.
Figures 15A and 15B respectively depict gain error and phase error vs. output power for the amplifier circuit comprising the components of Figures 1, 4, and 7.
Figures 16A and 16B respectively depict output power and PAE vs. frequency for the amplifier circuit comprising the components of Figures 1, 4, and 8.
Figure 17 depicts an exemplary transmitter application for the power amplifier disclosed herein.

DETAILED DESCRIPTION

The power amplifier disclosed herein comprises a single-stage amplifier that utilizes a cross-coupled tapped cascode topology together with a technique of applying an RF injection current into a wideband node to provide a single-stage power amplifier. In one embodiment, the power amplifier exhibits a 64% peak PAE, 29 dBm output power, and a 20.5 dB gain over a 2 GHz radio frequency band. Figure 1 shows a power amplifier circuit 100 configured to achieve these results. Amplifier circuit 100 comprises a cross-coupled cascode transistor unit 110 comprising a pair of cross-coupled cascode transistors 112, a cross-coupled switching transistor unit 120 comprising a pair of cross-coupled switching transistors 122, and an RF current generator 130. Broadly, the RF current generator 130 generates a differential RF injection current based on a differential input signal. Switching transistor unit 120 amplifies the injection current to generate an amplified injection current at the wideband node of the amplifier circuit, i.e., the source nodes of the cascode transistors 112. Cascode transistor unit 110 further amplifies the injection current to generate the desired amplified signal at the output 220 of the amplifier circuit 100, e.g., at the drain nodes of the cascode transistors 112. The amplitude of the output signal generally depends on the differential injection current and the supply voltage V_DD applied to the power amplifier circuit 100. As shown in Figure 1, a matching network 200 may be coupled to the output 220 of the amplifier circuit 100 to match the impedance of the amplifier circuit 100 to that of an external element, e.g., an antenna (not shown). While not required, it will be appreciated that the supply voltage V_DD may be applied to the power amplifier circuit 100 via the matching network 200, as shown in Figure 1. It will be appreciated that the transistors used to implement the power amplifier circuit 100 may comprise any known type of transistor, including but not limited to, NMOS, CMOS, BiCMOS, HBT, and III-V technology (including Bipolar and FET) transistors.
The improved operation of the amplifier circuit 100 relies on the cross-coupled configurations of the transistor units 110, 120 and the tap capacitors 114 of the cascode transistor unit 110. As shown in Figure 2 and explained in more detail herein, this configuration causes a signal B at the source node of a cascode transistor 112 to have the same amplitude as, but be out-of-phase from, both a signal A at the gate node of the same cascode transistor 112 and a signal C at a drain node of the opposite switching transistor 122. To further explain the details of the power amplifier circuit 100, the following description considers each of the RF current generator 130, switching transistor unit 120, and cascode transistor unit 110 separately.
RF current generator 130 generates a differential injection current I _RF+, I_RF- based on a differential input signal D ₊, D_-. Figures 3-6 depict various exemplary RF current generators 130. It will be appreciated, however, that current generators other than those shown herein may also be used.
The RF current generator 130 depicted in Figure 3 comprises a pair of injection transistors 132 configured to generate the RF differential injection current at the drain nodes from an input signal comprising an analog RF differential voltage signal applied to the gate nodes. In this embodiment, the injection transistors 132 are configured to operate as voltage-to-current converters. To provide the RF differential injection current to the rest of the amplifier circuit 100, the drain node of each injection transistor 132 couples to the drain node of the corresponding switching transistor 122 and to the source node of the corresponding cascode transistor 112.
An alternative RF current generator 130 may comprise a mixer configured to generate the RF differential injection current from an RF local oscillator signal and an input signal comprising a baseband differential input signal. In such an embodiment, upconversion to RF takes place inside the amplifier circuit 100, which removes the need for any upconversion outside the amplifier circuit 100. It will be appreciated that implementing upconversion inside the amplifier circuit 100 provides a more linear result, with respect to both upconversion and power amplification. Further, such mixers advantageously eliminate the need for separate RF drivers and other RF circuitry, which generally have high dynamic range requirements.
For example, Figure 4 depicts an RF current generator 130 comprising a transconductance mixer configured to generate the RF differential injection current at the RF current generator outputs from an RF local oscillator signal (LO) and an input signal comprising a baseband differential input current (BB). The RF current generator 130 of Figure 4 comprises a first pair of baseband transistors 134, a second pair of baseband transistors 135, and a pair of local oscillator transistors 136, e.g., NMOS transistors 136. The drain node of one of the local oscillator transistors 136 couples to a source nodes of corresponding transistors of the first and second pairs of baseband transistors 134, 135, and the drain node of the other local oscillator transistor 136 couples to the source nodes of the other of the first and second pairs of baseband transistors 134, 135, as depicted in Figure 4. The drains of the first pair of baseband transistors 134 also cross-couple with the drains of the second pair of baseband transistors 135. The positive input current signal D ₊ = BB ₊ is applied to the gate node of each of the first pair of baseband transistors 134 and the negative input current signal D _- = BB_- is applied to the gate node of each of the second pair of baseband transistors 135. The positive and negative local oscillator signals are applied to the gate nodes of respective transistors of the local oscillator transistor pair 136. As a result, the differential baseband input signal is upconverted to the LO frequency to generate the RF differential injection current I _RF+, I _RF- at the drain nodes of the baseband transistors 134, 135.
In another mixer example, the RF current generator 130 comprises a differential Quadrature mixer comprising an In-phase mixing unit 138 and a Quadrature-phase mixing unit 140, as depicted in Figure 5. The input signal comprises a baseband input signal having a differential In-phase portion BB _I+, BB _I- and a differential Quadrature-phase portion BB _Q+, BB _Q-, where the differential outputs from each mixing unit 138, 140 are cross-coupled to combine the In-phase and Quadrature-phase output portions to provide I _RF+ at one output and I _RF- at the other output. In-phase mixing unit 138 mixes the differential In-phase portion BB _I+, BB _I- with the differential In-phase local oscillator signal to generate the RF differential In-phase current I _{RF_I+}, I _{RF_I-}. Quadrature-phase mixing unit 140 mixes the differential Quadrature-phase portion BB _Q+, BB _Q- with the differential Quadrature-phase local oscillator signal to generate the RF differential Quadrature-phase current I _{RF_Q+}, I _{RF_Q-}. The differential outputs from each mixing unit 138, 140 are cross-coupled to combine the output positive In-phase and Quadrature-phase portions, I _{RF_I+} and I _{RF_Q+}, to provide I _RF+ at one output, and to combine the output negative In-phase and Quadrature-phase portions, I _{RF_I-} and I _{RF_Q-}, to provide I _RF- at the other output.
In still another mixer example depicted in Figure 6, the RF current generator 130 comprises an RF digital-to-analog converter/mixer. The input signal comprises a digital baseband input signal, and the RF current generator 130 upconverts the bits of the digital baseband input signal based on the an RF local oscillator signal to generate the RF differential injection current I _RF+, I _RF- . One exemplary RF digital-to-analog converter is disclosed in "A fully Digital Multimode Polar Transmitter Employing 17b RF DAC in 3G Mode" by Boos et al., and published at ISSCC 2011, Session 21, Cellular 21.7 (978-1-61284-302-5/11), which is incorporated herein by reference. This exemplary RF digital-to-analog converter employs 10 thermometer and 4 binary-coded bits with a high oversampling, using a GHz range clock, and providing overall 17b DAC resolution in 3G mode, and 19b in EDGE mode. It will be appreciated that other RF digital-to-analog converters may also be used.
Referring back to Figure 1, the cascode transistor unit 110 and switching transistor unit 120 are described. Cascode transistor unit 110 comprises a pair of cascode transistors 112 cross-coupled between the drain nodes and a pair of tap capacitors C _tap 114 at the gate nodes. The tap capacitors 114 are applied to the gate nodes to perform voltage division with the gate node capacitance (inherent in the gate node) to reduce the gate voltage swing. To that end, the tap capacitors 114 have a (matched) capacitance selected to control the signal from the cross-coupled drain nodes applied to the gate nodes, so that the amplitude of the gate node signal substantially equals the amplitude of the signal at the corresponding source node, or to differ from the amplitude of the source signal at the corresponding source node by no more than 50%. This voltage division protects the cascode transistors 112 from oxide overvoltage, reduces the capacitive loading of the matching network 200, and enables the desired amount of loop-gain to be provided for self-oscillation. The cross-coupling between the cascode transistors 112 flips the phase of the signals applied to the gate nodes, e.g., by 180°. As a result, the amplitudes at the gate and source nodes may be substantially equal, but the signals at these nodes are out of phase. This results in only half of the charges due to the parasitic capacitances C_par,_s being dissipated to ground, e.g., the signal ground, as compared to a classic cascode structure. Further, this configuration reduces the signal swing and the impedance of the source nodes of the cascode transistors 112, which relaxes both voltage stress and output power requirements of the switching transistors 122. Further still, using a cross-coupled tapped cascode transistor structure, as depicted in Figure 1, ensures that the swing at the gate node of the cascode transistors 112 tracks the output at the drain nodes over the entire bandwidth.
The switching transistor unit 120 comprises a pair of switching transistors 122 cross-coupled between the drain and gate nodes. The differential current output by the drain nodes of the cross-coupled switching transistors 122 increases the injection current I_RF , which leads to a wider locking range at a wideband node of the power amplifier circuit 100, i.e., the source nodes of the cascode transistors 112. Further, when the switching transistors 122 are on (e.g., when the output power exceeds some threshold) the majority of the injection current conducts through the cascode transistors 112. When the output power is low (e.g., below the threshold) the switching transistors 122 are off. When the RF current generator 130 comprises the injection transistors 132 of Figure 3, the cross-coupled configuration of the switching transistors enables the size of the injection transistors 132 to be reduced as long as the injection transistors 132 remain large enough so that the switching transistors 122 maintain a switching mode of operation.
The wideband impedance at the source node of the cascode transistors 112 combined with the switching properties of the switching transistors 122 (and in some cases, the injection transistors 132) provides square-wave current and voltage signals with steep edges at the source nodes of the cascode transistors 112. Because the voltage and current are not high simultaneously except during injection by the current generator 130, the resulting losses due to the switching transistor unit 120 and current generator 130 are dominated by the injection current output by RF current generator 130.
As shown in Figure 1, the power amplifier circuit 100 may be coupled to a matching network 200 configured to match the output impedance of the amplifier circuit 100 to that of an external load (not shown). Figures 7-10 depict exemplary matching networks 200. It will be appreciated that the present invention is not limited to the depicted matching networks 200.
Figure 7 depicts one exemplary matching network 200 coupling the load, represented by R_L, C_P, and L_P, and the reactive components represented by L_S, L_D, and C_S . L_S and C_S are connected as a series resonance circuit with the external load, which results in a negative (capacitive) reactance for low frequencies and a positive (inductive) reactance for high frequencies. L_D comprises a current source. At low frequencies L_D is a short, while at high frequencies, C_par,d (in the amplifier circuit 100) is a short. Thus, reactive components L_D and C_par,d set the limit for of the bandwidth over which the power amplifier can be efficient using the matching network 200 of Figure 7.
In some cases, it may be desirable to provide generally constant impedance to the drain nodes of the cascode transistors across a wide range of frequencies. As used herein, generally constant could mean actually constant, but it is more likely that the impedance has some variation, having for example no more than a 5% variation over the required or desired wideband frequency range. In other cases, a 1% might be required, while in still other cases 2%, 10%, 20%, or anything in between is acceptable. The acceptable variation might be affected by the specific components, values, or operating conditions of the circuit. Preferably, the limited variation applies over the entire required or desired wideband frequency range. Nevertheless, there may be frequency intervals where a larger variation is acceptable. The total length of such intervals preferably makes up no more than 20% of the required or desired wideband frequency range. In some cases, however, a total interval length relative to the wideband frequency range of 10%, 5%, 2%, 1%, or anything in between might be required. Figure 8 depicts an alternative higher order matching network 200 according to an embodiment configured to provide a generally constant impedance. In this embodiment, the matching network 200 couples a load represented by R_L, and reactive components represented by C_S, L_S, L_D, , L _D2, and C_D. In the matching network 200 of Figure 8, a first reactance unit comprises L _D1 and L _D2, which are serially coupled between signal ground, which may, e.g., correspond to the supply voltage V_DD, and each of the drain nodes of the cascode transistor unit 110. A second reactance unit comprises C_S and L_S serially coupled between a load resistor R_L and each of the drain nodes of the cascode transistor unit 110. A third reactance unit comprises at least one capacitor C_D coupled at one node between R_L and C_S , and at the other node between the inductors L _D1 and L _D2 of the first reactance unit. The inductors of the first reactance unit generate a first high positive reactance at low frequencies, which keeps the losses low, even for low frequencies, where a single inductor is considered to be a short, and a second positive reactance at high frequencies. The second reactance unit generates a negative reactance at low frequencies and third high positive reactance at high frequencies. As the operating frequency increases, C_par,d (in the amplifier circuit 100) goes towards a short, which leads to increased power consumption. By decreasing the reactance between the drain nodes of the cascode transistors 112, e.g., by using the capacitor(s) C_D of the third reactance unit, which short at higher frequencies, the effect of C_par,d decreases (it is effectively cancelled). Such reactance properties yield higher impedances at the drains of the cascode transistors 112, and therefore, widen the bandwidth and efficiency with respect to frequency.
Figures 9 and 10 depict other exemplary matching networks 200 comprising a transformer 230. In Figure 9, the transformer 230 is configured as a balun, where one side of the transformer is differentially connected to the power amplifier circuit 100 and the other side connects to the load R_L in a single-ended manner. In Figure 10, one side of the transformer is differentially connected to the power amplifier circuit 100 while the other side differentially connects to the load R_L . The performance of the matching networks 200 in Figures 9 and 10 is generally the same, and is generally comparable to that of the matching networks 200 of Figures 7 and 8. In addition, because the matching network 200 of Figure 10 is not a balun, an additional balun would be needed before the antenna for this embodiment.
Figures 11-16 depict various simulation results for exemplary cases of the power amplifier disclosed herein. For example, Figures 11-14 depict parameter and performance results for a power amplifier configuration comprising the power amplifier circuit 100 of Figure 1, the RF current generator 130 of Figure 3, and the matching network 200 of Figure 7. Figures 11A and 11B respectively show the supply voltage (V_DD ) and cascode bias voltage (B_cc ) with respect to input power. For lower output power levels (linear operating mode), the supply voltage was kept constant at 0.48 V. As the input power increased, the supply voltage was increased to allow more headroom for the output signal. As depicted in Figure 11A, the supply voltage is 3.0 V at the peak input power. Figure 11B shows that the cascode bias voltage has similar characteristics to the supply voltage, except that the values go from 0.865 V for the linear mode of operation to a maximum bias of 1.9 V.
Figure 12 depicts the output power, gain, and PAE results as a function of the input power at 2 GHz. As shown in Figure 12, the output power tracks the input power linearly, which is also demonstrated by the flat gain response. The PAE peaks at 64%. When the input power is backed off by 16 dB, the PAE reduces to 37%. Considering the power range, Figure 12 demonstrates that the power amplifier operates over more than an 80 dB range, which satisfies WCDMA requirements.
Figures 13A and 13B depict the frequency response for the output power and PAE for different input powers. At peak output power, Figure 13A demonstrates that the -3 dB bandwidth of the power amplifier circuit 100 is 1.2 GHz (between 1.4 and 2.6 GHz). Figure 13A further demonstrates that the bandwidth increases for lower input powers. Further, Figure 13B demonstrates that the PAE at the peak output power is above 50% between 1.6 GHz and 2.6 GHz. When the output power is backed off by 18 dB, the PAE is at a constant value of 20% for in band frequencies, and rolls off quickly outside the bandwidth.
Lastly, Figures 14A and 14B depict the linearity of the gain error and phase error, which was measured in a static fashion over the WCDMA output power range of approximately 80 dB. As shown in Figure 14A, the power amplifier circuit 100 has a gain error of 0.2 dB over the entire range and linear gain during class AB operations. For the AM-PM conversion (Figure 14B), which represents how much the phase changes when the amplitude changes, the total phase error is 17° over the entire output power range. The total error (both AM-AM and AM-PM) in the power amplifier is within a range of pre-distortable values.
Figures 15A and 15B depict linearity performance results for the power amplifier comprising the power amplifier circuit 100 of Figure 1, the RF current generator 130 of Figure 4, and the matching network 200 of Figure 7. Figure 15A demonstrates that the gain error degrades slightly when the transconductance amplifier replaces the injection transistor embodiment of the RF current generator 130, but the total phase error improves to roughly 1° (Figure 15B). Again, the total error (both AM-AM and AM-PM) in the power amplifier is within pre-distortable values.
Figures 16A and 16B depict the frequency response for the output power and PAE for the power amplifier comprising the power amplifier circuit 100 of Figure 1, the RF current generator 130 of Figure 4, and the matching network 200 of Figure 8. At peak output power, Figure 16A demonstrates that the -3 dB bandwidth of the power amplifier circuit 100 is 2.0 GHz (between 0.6 and 3.6 GHz). Figure 16A further demonstrates that the bandwidth stays the same or increases for lower input powers. Further, Figure 16B demonstrates that the PAE at the peak output power is above 50% between 0.6 GHz and 3.4 GHz. When the output power is backed off by 18 dB, the PAE is at a constant value of approximately 20% for frequencies between 0.7 and 2.5 GHz.
The power amplifier circuit disclosed herein comprises an RF switched-mode power amplifier (SMPA). In an RF SMPA, switching typically dominates the losses, e.g., from power dissipation due to charging and discharging capacitances between the supply and ground. By using a tuned circuit, e.g., an LC-oscillator, the reactive energy may alternate between capacitors and inductors in the matching network 200, rather than all of it being dissipated in the switch resistances every RF cycle. Some losses remain, however, due to losses of inductors and capacitors, and due to the non-zero currents and voltages in both the cascode transistors 112 and the switching transistors 122. In order to reduce these losses, the capacitance may be minimized, resulting in a low tank Q when connected to the resistive output load, and a wide bandwidth. For example, by integrating the cascode capacitances into the matching network 200, as shown in Figures 1 and 7 or 8, the losses can be reduced.
The power amplifier circuit 100 disclosed herein may be implemented in the STMicroelectronics 65 nm CMOS process with eight metal layers and MIM capacitors. In one exemplary case, the cascode transistors 112 may be implemented using thick oxide 2.5 V I/O devices. In one exemplary layout, all transistors are drawn in a common-centroid layout to minimize mismatch. Because the area of each individual transistor is also large, the resulting mismatch is small. In one implementation, the resulting chip has a chip area of 0.52x0.48 mm², including pads.
It will be appreciated that the power amplifier circuit 100 disclosed herein provides a single-stage amplifier solution using 65 nm CMOS to achieve, in one embodiment, a 2 GHz bandwidth, 29 dBm output power, 20.5 dB gain, and 64% PAE. Such performance results are as good as some past multiple stage solutions, and generally are better than most past amplifier solutions.
While not required, the power amplifier 100 and matching network 200 disclosed herein may be used in a Hybrid-Envelope Elimination and Restoration (H-EER) system, such as shown in Figure 17, where the amplifier operates as a Mixed-Mode Power Amplifier (MMPA). In this example, MMPA refers to the operation of the amplifier as a self-oscillating Switched-Mode Power Amplifier (SMPA) for high output power levels, and as a linear class AB power amplifier for low output power levels, e.g., when the power amplifier is not the main power consumer. This increases the power range of the system, overcoming one of the key issues of SMPAs, which is their poor power range. Such improvements are important for WCDMA operation, which has a power control range of 80 dB.
The power amplifier circuit disclosed herein utilizes a cross-coupled tapped cascode topology together with a technique of applying an RF injection current into a wideband node to provide a single-stage power amplifier with improved PAE, output power, and gain over a wide radio frequency band. It will be appreciated that while the power amplifier disclosed herein may be used for wideband operations, the power amplifier also provides high efficiency, output power, and gain for narrower frequency band applications.
The power amplifier circuit comprises a current generator, a pair of cross-coupled switching transistors, a pair of cross-coupled cascode transistors, and first and second tap capacitors. The current generator is configured to generate a radio frequency differential injection current at a differential current generator output based on an input signal. The input signal may comprise any one of an analog input voltage signal at the radio frequency, or a baseband input signal, e.g., an analog or digital baseband input signal. The switching transistors each include a source node, a gate node, and a drain node, and are cross-coupled such that the drain node of one switching transistor couples to the gate node of the other switching transistor. Further, the drain nodes of the switching transistors are coupled to the differential current generator output to receive the differential injection current. The cascode transistors each include a source node, a gate node, and a drain node, and are cross-coupled such that the drain node of one cascode transistor couples to the gate node of the other cascode transistor. The cross-coupled cascode transistors are configured to generate a differential amplified output signal at the drain nodes of the cascode transistors based on a supply voltage operatively coupled to the drain node of the cascode transistors and the differential injection current applied to the source node of the cascode transistors. The first and second tap capacitors couple to the respective gate nodes of the cascode transistors. In one embodiment, the tap capacitors have a capacitance selected to control the amplitude of a gate signal at the gate nodes of the cascode transistors to substantially equal the amplitude of a source signal at the corresponding source nodes of the cascode transistors. As used herein, substantially equal could mean equal, but more likely means that the amplitude of the gate signals at the gate nodes of the cascode transistors has some deviation, e.g., deviates within a certain range around the amplitude of the source signals at the corresponding source nodes of the cascode transistors. The deviation range might, for example, be 10%. In other embodiments, the deviation range might be 5% in some cases, 2% in other cases, 20% in other cases, 50% in other cases, or any value in between, depending on the specific components values or operating conditions of the circuit. Preferably, this deviation range is over the entire dynamic range of the amplifier circuit. Nevertheless, there may be intervals where the deviation range is larger. The total length of such intervals preferably makes up no more than 20% of the entire dynamic range, but in some cases, a total interval length of, e.g., 10% or 5% or 2% might be required.
The present invention may, of course, be carried out in other ways than those specifically set forth herein without departing from essential characteristics of the invention. The present embodiments are to be considered in all respects as illustrative and not restrictive, and all changes coming within the meaning of the appended claims are intended to be embraced therein.

Claims

A matching network for matching an output impedance of an amplifier circuit to an external load resistor (R_L), said matching network comprising:
two first reactance units each comprising two inductors (L_D1, L_D2) adapted to be serially couplable between a signal ground and respective first and second outputs of the amplifier circuit for generating a first positive reactance at low frequencies and a second positive reactance at high frequencies;

two second reactance units each comprising at least one series capacitor (C_s) and at least one series inductor (L_s) adapted to be serially couplable between a respective end of the external load resistor (R_L) and the respective first and second outputs of the amplifier circuit, for generating a negative reactance at low frequencies and a third positive reactance at high frequencies; and

two third reactance units each comprising at least one control capacitor (C_D) adapted to be couplable at one node between the respective end of the external load resistor (R_L) and the respective end of the second reactance unit and at the other node between the inductors (L_D1, L_D2) of the respective first reactance unit, for generating a short at high frequencies so as to reduce a parasitic capacitance at the first and second outputs of the amplifier circuit at high frequencies,

wherein a generally constant impedance is provided across a wideband frequency range.
The matching network of claim 1 wherein the signal ground corresponds to a supply voltage.
An arrangement comprising the matching network of claims 1 or 2 and the external load resistor (R_L).